https://twitter.com/allixender/status/1323293939865104384
the layer E_203_vooluveekogu_j (lit. translation: flowing water bodies as lines) from the file ETAK_EESTI_GPKG.gpkg, downloaded in Geopackage format (GPKG) ( 1.2G ) under Estonian Landboard Open Data License.
I am using the column tyyp which denotes the “magnitude” of the river, i.e.: 50=river, 40=stream … channel, 10=ditch. The higher the values, the more intense the colour is meant to shine.
The Jupyter notebook (view here) is provided in my GitHub repo.
Geopandas for reading the Data and pre-processing
Datashader for fancy glow plotting more than 845 000 geometries within a few seconds
What Datashader does, is resampling the hundreds of thousands geometries into a much smaller number of pixels. It is very efficient and optimized, using things like Dask and Numba under the hood. Datashader itself exposes a comparatively small API. However, I struggled to integrate the plotting with the typical elements I am used from e.g. matplotlib , such as title, legend and geographic coordinates.
Suggestions welcome.